A Perspective

Michael E. Himmel, William S. Adney, John O. Baker, Richard Elander,
James D. McMillan, Rafael A. Nieves, John J. Sheehan,

Steven R. Thomas, Todd B. Vinzant, and Min Zhang

Biotechnology Center for Fuels and Chemicals, National Renewable
Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401

Conversion of the fermentable sugars residing in lignocellulosic waste and energy crops can conservatively yield approximately 100 billion gallons of fuel-grade ethanol per year in the United States alone. However, the cellulosic biomass-to-alcohol bioconversion process must be proven economical before industry can commercialize this technology. The U. S. Department of Energy, Office of Fuels Development supports a program to develop a commercially viable process for producing ethanol transportation fuel from renewable biomass resources. Bioconversion technologies developed to date take advantage of a diverse array of pretreatments, followed by enzymatic or chemical saccharification, and fermentation. Progress during recent years in pretreatment and fermentation technologies promises to significantly improve overall process economics. Examples include the development of two-stage dilute acid, nitric acid/mechanical disruption, and countercurrent percolation pretreatments. Recently developed, genetically engineered ethanologens also promise to improve process economics and include Escherichia coli, Saccharomyces sp., and Zymomonas mobilis. These microorganisms reportedly ferment xylose and glucose mixtures with high efficiency. This review presents an up-to-date picture of the advanced technology aspects of unit operations key to successful biomass-to-ethanol processing plants with speculation about future focus in this field by NREL researchers.

Processes for production of fuel-ethanol from lignocellulosic materials involving micro-organisms and biomass are very complex. Therefore, it can be difficult to design a full-scale or even a pilot-plant facility based on lab-scale data only. One of the most efficient ways of assessing the technological and economic feasibility of such processes is through the use of techno-economic computer modelling. Since a biomass-to-ethanol process consists of several process steps, all strongly interdependent, it is extremely difficult to identify the relative merits of a change in

To distillation

Figure 9. Schematic flowsheet of evaporation with stripper incorporated.

one subprocess and its influence on the final production cost of ethanol. By computer simulations different process strategies can be evaluated and experiments in a bench — scale unit can serve as feed-back to the process simulations.

A bench-scale unit is due to its flexibility very well suited for studying various process configurations and also the recycling of process streams. In this chapter we have shown how the use of a bench-scale unit in combination with process simulations can be used to study process integration. In an environmentally sustainable process, waste water must be minimised. Several process configurations with recycling of process streams were investigated which will result in reduced waste water streams as well as reduced energy consumption. The experimental investigation in the bench-scale unit showed that recycling of non-volatile compounds had an inhibitory effect on the fermentation process. This suggests an incorporation of a multi-effect evaporation line in the process. If internal energy integration is employed, the demand for extra energy can be reduced. The data thus gathered will now be included in Aspen Plus to refine the models which will lead to new simulation results. The aim is to reach an optimal process for production of ethanol which is economic and environmentally feasible. Final process optimisation should preferably be performed on a pilot-scale, since it is difficult to study the whole effect of energy integration options in a bench-scale unit.

Acknowledgments

The Swedish National Board for Industrial and Technical Development is gratefully acknowledged for its financial support.

The major use of lactic acid is in food and food-related applications, which, in the U. S., accounts for approximately 85% of the demand. The rest (-15%) of the uses are for nonfood industrial applications. As a food acidulant, lactic acid has a mild acidic taste in contrast to other food acids. Lactic acid is nonvolatile, odorless, and is classified as GRAS (generally recognized as safe) for use as a general purpose food additive by the FDA in the U. S. and other regulatory agencies elsewhere. It is a very good preservative and pickling agent for sauerkraut, olives, and pickled vegetables. It is used as acidulant/flavoring/pH buffering agent or inhibitor of bacterial spoilage in a wide variety of processed foods, such as candy, breads and bakery products, soft drinks, soups, sherbets, dairy products, beer, jams and jellies, mayonnaise, and processed eggs — often in conjunction with other acidulants (6). An emeiging new use for lactic acid or its salts is in the disinfection and packaging of carcasses, particularly those of poultry and fish, where the addition of aqueous solutions of lactic acid and its salts during the processing increased shelf life and reduced the growth of anaerobic spoilage organisms such as Clostridium botulinum (7-8).

A large fraction (> 50%) of the lactic acid for food-related uses goes to produce emulsifying agents used in foods — particularly for bakery goods. These emulsifying agents are esters of lactate salts with longer chain fatty acids, and the four important products are calcium and sodium, stearoyl-2-lactylate, glyceryl lactostearate, and glyceryl lactopalmitate. Of the stearoyl lactylates, the calcium salt is a very good dough conditioner, and the sodium salt is both a conditioner and an emulsifier for yeast-leavened bakery products. The glycerates and palmitates are used in prepared cake mixes, other bakery products, and in liquid shortenings. In prepared cake mixes, the palmitate improves cake texture, whereas the stearate increases cake volume and permits mixing tolerances (6). The manufacture of these emulsifiers requires heat-stable lactic acid — hence, only the synthetic or the heat-stable fermentation grades are used for this application.

Technical-grade lactic acid has long been in use in the leather tanning industry as an acidulant for deliming hides and in vegetable tanning. In various textile finishing operations and acid dying of wool, technical-grade lactic acid was used extensively. Cheaper inorganic acids are now more commonly used in these applications. The future availability of lower cost lactic acid and the increasing environmental restrictions on waste salt disposal may reopen these markets for lactic acid.

Lactic acid is currently used in a wide variety of small-scale, specialized industrial applications where the functional specialty of the molecule is desirable. Some examples are pH adjustment of hardening baths for cellophane that is used in food packaging, terminating agent for phenol-formaldehyde resins, alkyd resin modifier, solder flux, lithographic and textile printing developers, adhesive formulations, electroplating and electropolishing baths, detergent builders (with maleic anhydride to form carboxymethoxysuccinic acid-type compounds). Because of the current high cost and low volume of production, these applications account for only 5-10% of the consumption of lactic acid (6, 9).

Lactic acid and ethyl lactate have long been used in pharmaceutical and cosmetic applications and formulations, particularly in topical ointments, lotions, parenteral solutions, and biodegradable polymers for medical applications (such as surgical sutures, controlled-release drugs, and prostheses). A substantial part of pharmaceutical lactic acid is used as the sodium salt for parenteral and dialysis applications. The calcium salt is widely used for calcium-deficiency therapy and as

an effective anti-caries agent. As humectants in cosmetic applications, the lactates are often superior to natural products and more effective than polyols (6, 9). Ethyl lactate is the active ingredient in many anti-acne preparations. The use of the chirality of lactic acid for synthesis of drugs and agrichemicals is an opportunity for new applications for optically active lactic acid or its esters. The chiral synthesis routes to R (+) phenoxypropionic acid and its derivatives using S (-) lactate ester as a chiral synthon has been described (6). These compounds are used in herbicide production. Another use as an optically active liquid crystal whereby lactic acid is used as a chiral synthon has been recently described (10). These advances could open new small-volume specialty chemical opportunities for optically active lactic acid and its derivatives.

Lactic acid can be manufactured by either (1) chemical synthesis or (2) carbohydrate fermentation — both are used for commercial production. In the U. S., lactic acid is manufactured synthetically by means of the lactonitrile route by Sterling Chemicals, Inc. In Japan, Musashino Chemical Co. used this technology for all of Japan’s production. CCA Biochemical b. v. of the Netherlands uses carbohydrate fermentation technology in plants in Europe and Brazil and markets worldwide. Prior to 1991, the annual U. S. consumption of lactic acid was estimated at 18,500 metric tonnes, with domestic production of approximately 8,600 tonnes, by Sterling Chemical and the rest imported from Europe and Brazil. The worldwide consumption was estimated at approximately 40,000 tonnes/yr.

(1) Chemical Synthesis. The chemical-synthesis routes produce only the racemic lactic acid. The commercial process is based on lactonitrile, which used to be a by-product from acrylonitrile synthesis. It involves base catalyzed addition of hydrogen cyanide to acetaldehyde to produce lactonitrile. This is a liquid-phase reaction and occurs at atmospheric pressures. The crude lactonitrile is then recovered and purified by distillation and is hydrolyzed to lactic acid by using either concentrated hydrochloric or sulfuric acid, producing the corresponding ammonium salt as a by-product. This crude lactic acid is esterified with methanol, producing (1) methyl lactate, which is recovered and purified by distillation and hydrolyzed by water under acid catalysts to produce lactic acid, which is further concentrated, purified, and shipped under different product classifications, and (2) methanol, which is recycled (equations 1-3).

СНз CHO + HCN catalyst > CH3 CHO HCN (1)

CH3 CHO HCN + H20 + і H2 S04 ->• CH3 CHOH COOH + і (NH4>2 SO4 (2) CH3 CHOH COOH + CH3 OH -> CH3 CHOH COOCH31 + h2o

CH3 CHOH COO CH3 + H2O -> CH3 CHOH COOH + CH3OH t (3)

Other possible chemical synthesis routes for lactic acid include base catalyzed degradation of sugars; oxidation of propylene glycol; reaction of acetaldehyde, carbon monoxide, and water at elevated temperatures and pressures; hydrolysis of chloropropionic acid (prepared by chlorination of propionic acid), and nitric acid oxidation of propylene, among others. None of these routes have led to technically and economically viable processes (9, 11).

(2) Carbohydrate Fermentation. The fermentation technology can make a desired stereoisomer of lactic acid. The existing commercial production processes use homolactic organisms, such as Lactobacillus delbrueckii, L. bulgaricus,

L. leichmanii. A wide variety of carbohydrate sources can be used (molasses, com syrup, whey, dextrose, cane, or beet sugar). The use of a specific carbohydrate feedstock depends on its price, availability, and purity. Proteinaceous and other complex nutrients required by the organisms are provided by com steep liquor, yeast extract, and soy hydrolysate, for example. Excess calcium carbonate is added to the fermenters to neutralize the acid produced and produce a calcium salt of the acid in the broth. The fermentation is conducted as a batch process, requiring 4 to 6 days to complete. Lactate yields of approximately 90% (w/w) from a dextrose equivalent of carbohydrate are obtained. Keeping the calcium lactate in solution is desirable so that it can be easily separated from the cell biomass and other insolubles, and this limits the concentration of carbohydrates that can be fed in the fermentation and the concentration of lactate in the fermentation broth, which is usually around 10% (w/v). The broth containing calcium lactate is filtered to remove cells, carbon treated, evaporated, and acidified with sulfuric acid to convert the salt into lactic acid and insoluble calcium sulfate, which is removed by filtration. The filtrate is further purified by carbon columns and ion exchange and evaporated to produce technical and food-grade lactic acid, but not a heat-stable product, which is required for the stearoyl lactylates, polymers, and other value-added applications. The technical-grade lactic acid can be esterified with methanol or ethanol, and the ester is recovered by distillation, hydrolyzed by water, evaporated, and the alcohol is recycled. This separation process produces a highly pure product, which, like the synthetic product, is water white and heat stable (equations 4-7).

C6 Hi2 Об + Ca(OH)2 Fermentation > (2 . CH3 CHOH COO) CA++ + 2H20 (4)

(2 • CH3 CHOH COO) Ca++ + H2S04—— > 2 CH3 CHOH COOH + CaS04 і (5)

CH3 CHOH COOH + CH3 OH—— > CH3 CHOH COO CH3 t + H20 (6)

CH3 CHOH COOCH3 + H20——- > CH3 CHOH COOH + CH3 OH T (7)

Some of the major economic hurdles and process cost centers of this conventional carbohydrate fermentation process are in the complex separation steps that are needed to recover and purify the product from the crude fermentation broths. Furthermore, approximately one ton of gypsum by-product is produced and needs to be disposed of for every ton of lactic acid produced by the conventional fermentation and recovery process. These factors had made large-scale production by this conventional route economically and ecologically unattractive.

Lignin is a long chain heterogeneous polymer composed largely of phenylpropane units most commonly linked by ether bonds. It effectively protects the woody plants against microbial attack and only a few organisms including rot-fungi and some bacteria can degrade it (57). The conversion of cellulose and hemicellulose to fuels and chemicals will generate lignin as a by-product that can be burned to provide heat and electricity, converted to low-molecular weight chemicals, and used in the manufacture of various polymeric materials. As lignin makes up 15-25% in some lignocellulosic biomass, the selling price of lignin has a very large impact on ethanol price (18).

In recent years, removal of lignin from lignin-carbohydrate complex (LCC) has received much attention because of potential application in pulp and paper industry. The lignin barrier can be disrupted by a variety of pretreatments rendering the cellulose and hemicellulose more susceptible to enzymatic attack (52). There are many papers about microbial breakdowns of lignin, the enzymes and the pathways (55-56). The degradation of lignin by the basidiomycete Phanerochaete chrysosporium is catalyzed by extracellular peroxidases (lignin peroxidase, Lip and manganese peroxidase, MnP) in a H202-dependent process (57, 58). However, due to extreme complexity of the problem, a vast amount of research needs to be done to understand all the factors involved in lignin biodegradation process (59).

Xiaoyan Xiong, William S. Adney, Todd B. Vinzant, Yat-Chen Chou,
Michael E. Himmel1, and Steven R. Thomas

Biotechnology Center for Fuels and Chemicals, National Renewable
Energy Laboratory, 1617 Cole Boulevard, Golden, CO 80401

Active, thermostable Microbispora bispora Bgl В (i-D-glucosidase was expressed in Streptomyces lividans TK24 cells transformed with plasmid pIJ702 carrying the bglB coding sequence under the control of Streptomyces longisporus STI-II trypsin inhibitor promoter. The recombinant enzyme, 5/Bgl B, has a molecular weight of 54 kDa, an isoelectric pH of 5.0, shows resistance to glucose inhibition, and is optimally active on o-nitrophenyl р-D-glucopyranoside at 57°C. This recombinant (i-D-glucosidase was more active on aryl-glycosides than on cellobiose. We also report a successful mutagenesis strategy used to achieve increased levels of 5/Bgl В expression in this host organism. Screening mutants created by low fidelity PCR using Taq polymerase in the presence of manganese ion revealed a series of up-regulated clones, one yielding 235 mg/L of 5/Bgl B.

p-D-glucoside glucohydrolases (EC 3.2.1.21), or p-D-glucosidases, catalyze the hydrolysis of O-glycosyl bonds in aryl — and alkyl-glucosides, as well as in many p-linked disaccharides and some oligosaccharides. These enzymes are produced by plants, animals, and most known microbiota and have been extensively reviewed (7,2). Each of these enzymes displays a distinct pattern of relative activity on an array of p-glucosides. A subset of p-D — glucosidases are especially proficient at the hydrolysis of cellobiose, and are often referred to as cellobiases.

In general, thermotolerant enzymes have higher turnover rates and better tolerate the stresses of use in large-scale processes. Thus, they are of interest to industry (5). Thermophilic P-D-glucosidases have been found in Clostridium stercorarium, T^ = 65°C

(4) ; Clostridium thermocellum NQB 10682, T^ = 65°C (5); Thermomonospora sp. strain YX, T^ = 55°C (6); and Acidothermus cellulolyticus, T^ = 75°C (7). Caldophilic P-D — glucosidases have also been isolated from Caldocellum saccharolyticum, Topt = 85°C (5);

Corresponding author

© 1997 American Chemical Society

Clostridium thermocopriae JT3-3, = 80°C (9); and Thermus strain Z-l, Topt = 85°C

m.

Typically, native strains are not prolific producers of p-D-glucosidase activity, so researchers have used recombinant DNA expression strategies to obtain reasonable quantities of these enzymes. Genetically engineered expression from E. coli of various thermostable P-D-glucosidases including C. saccharolyticum (5), C. thermocellum NUB 10682 (77), C. thermocellum NOB 10682 (72), and Rhodothermus marinus (13) has been demonstrated in several laboratories using lambda phage or plasmid vectors in E. coli and other host bacteria.

In 1986, Waldron and со workers reported the isolation of a new thermo tolerant actinomycete, Microbispora bispora, from warm compost (14). This microorganism produced thermostable p-D-glucosidase activity that was resistant to inhibition by glucose concentrations as high as 30% w/v. Upon screening a genomic library of M. bispora DNA, Wright (75) discovered two DNA fragments of 4.0 kb and 2.1 kb coding for two distinct p-D-glucosidases, Bgl A and Bgl B, respectively. The bglB gene encodes M. bispora p-D-glucosidase B, MBgl B, which is highly resistant to feedback inhibition by glucose, but is expressed only very weakly from its native promoter in E. coli. We report the cloning of the coding sequence for MBgl В under the control of the STI-II trypsin inhibitor promoter from Streptomyces longisporus (16) in plasmid pIJ702, as well as the expression and characterization of active, thermostable enzyme from transformed Streptomyces lividans TK24, S/Bgl B. We also report on our efforts to mutagenize the STI-II promoter to achieve higher levels of 5/Bgl В expression in this host organism.

It is still an open question if the enterobacteria of the genera Klebsiella and Citrobacter or the Clostridia are more suitable for a 1,3-PD production process. From the viewpoint of an interested company there will probably be a preference for the clostridia as both enterobacterial species are classified as opportunistic pathogens and thus would require costly safety precautions. On the other hand, as shown in the preceding section the strains of Clostridium butyricum which are presently in use cannot entirely compete with strains like Klebsiella pneumoniae DSM 2026 at least if productivity is concerned (38). Several efforts have been made in the last five years to meet this shortcoming. The medium has been improved, better culture techniques have been elaborated, new strains have been isolated from nature or selected from culture collections, and mutants have been obtained which are considerably increased in product tolerance and yield.

Medium and Culture Conditions. For estimations of the performance of Klebsiella and Clostridium cultures virtually different media had been in use. When the medium used for Klebsiella (14) was applied to Clostridium butyricum DSM 5431 indeed the steady state product concentration in continuous cultures could be increased. Higher iron con­tent and the presence of citrate as a complexing agent were found to be the main cause of this stimulation (24). The yeast extract formerly used in a concentration of 1 g/1 (17) could be replaced by biotin with only a slight loss in productivity.

Continuous culture is a good tool for anaerobic glycerol fermentation particularly when a second stage is used to increase the final product concentration (8). However, for maximum propanediol content and simple operation batch cultures appear to be mo­re advantageous. Giinzel et al. (17) described a fedbatch culture that consumed about 110 g/1 of glycerol supplied in three successive additions over a period of 24 h which resulted in a 1,3-propanediol concentration of 56 g/1. Recently this process was automa­ted using the pH decrease as a signal for glycerol addition. The culture was kept at a slight glycerol excess so that substrate limiting intervals were avoided; the residual gly­cerol was used up towards the end of the fermentation after turning off the glycerol supply. The results were about the same as in discontinuous feeding (24a). A similar system was described by Saint-Amans et al. (27) for C. butyricum VPI 3266 using C02 production for control of glycerol supply. Probably due to the properties of the particu­lar strain (see below) the final product concentration was higher (65 g/1), but the fer­mentation time was about three times as long.

Procedures to increase the productivity by immobilization or cell recycling have been rarely elaborated for the glycerol fermentation. Pflugmacher and Gottschalk (23) described a fixed bed loop reactor culture of Citrobacter freundii using polyurethane foam as carrier substance. In comparison to a stirred tank reactor culture (8) the propa­nediol productivity was more than doubled, but the propanediol concentration could not be increased beyond 19 g/1. Similar results were obtained with Clostridium butyri — cum using a crossflow filtration module. Although it was possible to increase the pro­ductivity up to 3 times of the conventional continuous culture, the steady state propane­diol concentration was not higher than 14 g/1 (Reimann and Biebl, unpublished results).

Use of Cosubstrates. As pointed out above a 1,3-propanediol yield of 72 mol per 100 mol of glycerol cannot be surpassed. If however another fermentable electron donor substrate could be applied in addition to glycerol, a 100 % product yield is conceivable. In the patent literature procedures are described in which glucose is the cosubstrate and enterobacteria the converting organisms. It is proposed to grow the cultures initially with glycerol alone to induce the 1,3-propanediol forming enzymes. Glucose is added either directly to the growing culture (32) or used in a mixture with glycerol in a resting-cell culture without an ammonium source (16). 91 to 100 % of the glycerol were converted to 1,3-propanediol by this technique.

If glucose is considered as cosubstrate it has to be kept in mind that hexoses are less reduced substrates than glycerol, so that twice as much glucose than glycerol is needed on a weight basis to provide the same amount of reducing equivalents. This im­plies that use of glucose in 1,3-propanediol production would only be useful if glucose is available at a considerably lower price than glycerol, which appears to be presently the case, especially in the USA.

Another prerequisite for glucose addition should be that it is converted mainly to acetate in the presence of glycerol and not to ethanol or butyrate. This has been verified for C. butyricum (6). However, there are other physiological barriers in simultaneous metabolization of glycerol and glucose. As reported by Biebl and Marten (6) glucose is fermented much slower than glycerol by glycerol-fermenting Clostridia so that a com­plete conversion of the glycerol cannot be obtained. A fermentable substance other than glucose which is a cheaply available chemical bulk product, ethylenglycol, proved to be unsuitable as a cosubstrate. In the presence of glycerol this diol did not release reducing equivalents in the course of acetate formation as expected, but utilized hydrogen equi­valents to yield ethanol as product, while glycerol was oxidized to acetate.

New Strains of Clostridium butyricum. Whereas in the case of Klebsiella pneumoniae the type strain of the species is still unsurpassed in glycerol fermentation, in the case of Clostridium butyricum new screening efforts have led to strains that are markedly im­proved in product tolerance (22). In fed batch culture the best strain (E5) produced up to 66 g/1 of propanediol from 122 g/1 of glycerol compared to 55 g/1 from 108 g/1 of glyce­rol with strain DSM 5431 (5,17). The new strain was distinguished by very low hydro­gen evolution (24a). However, its maximum growth rate and productivity were distinct­ly lower. The strain used by Saint-Amans et al. (26), C. butyricum VPI 3266, is proba­bly closely related.

Genetic Approaches to Strain Improvement in C. butyricum have been undertaken only very recently (22). NTG mutations of strain DSM 5431 were selected in the pre­sence of high propanediol concentrations and further on bromide-bromate glucose me­dium to obtain mutants that sustained substantially higher product concentrations and were strongly reduced in hydrogen evolution (Reimann and Biebl, unpublished data). In fedbatch culture the best of them was able to convert up to 130 g of glycerol to about 70 g of 1,3-propanediol. Thus it resembles the new isolates obtained by the same group as well as the VPI strain used by Saint-Amans et al. (27,28). If these strains will be sub­jected to genetic improvement further increase in product concentration can be expec­ted.

Attempts to generate a strain from the product tolerant, hydrogen reduced mutants that was defect or affected in butyrate formation failed with several methods. In contrast it was possible, to obtain butyrate reduced mutants from the isolate E5 by selection on allylalcohol (22). These mutants were not much changed in propanediol production but exhibited a hydrogen production that was near the physiological maximum whereas the parent strain was very low in hydrogen evolution. This result in addition to fermenta­tion data and experiments that showed stimulation of 1,3-propanediol production in the presence of an aldehyde prompted the authors to assume that glycerol dehydration the product of which is the toxic 3-hydroxypropionaldehyde is the rate limiting step. Con­sequently, a certain amount of reducing equivalents has to be disposed via butyrate and/or hydrogen, as the 1,3-propanediol pathway can be only varied within a narrow range. A possibility to claim all reducing equivalents produced in glycolysis for 1,3- propanediol formation would be to overexpress the glycerol dehydratase along with the

1.3- propandiol dehydrogenase using recombinant DNA techniques. Such experiments are presently in progress.

Genetic Improvement with Enterobacterial Strains. From Citrobacter and Klebsiella strains over-producing mutants have not been isolated. On the other hand basic genetic research on the key enzymes of the process is more advanced than in Clostridia. In both species the genes of the dha regulon which encode for the glycerol dehydrogenase, the dihydroxyacetone kinase, the gycerol dehydratase and the 1,3-propanediol dehydroge­nase have been cloned and expressed in E. coli (10,11,30,31). A new construction of the dha regulon has been achieved by Cameron (personal communication). The amounts of

1.3- propanediol wich have been reached in the recombinant E. coli strains do not yet approach that of the original strains (i. e. 9 g/1 for strains with Citrobacter and 6 g/1 for strains with the Klebsiella genes), but in the long term highly productive recombinant Klebsiella strains can be expected as soon as the optimized genes have been reintrodu­ced into the donor strains.

Daniel W. Karl1, Kris M. Roth2’3, Frederick J. Schendel4, Van D. Gooch5,

and Bruce J. Jordan2

Daniel Karl Scientific Consulting, 430 Saratoga Street South,

St. Paul, MN 55105

2Morris Ag-Energy, Inc., P. O. Box 111, Morris, MN 56267
department of Chemistry, University of Minnesota, Morris, MN 56267
4ENCORE Technologies, 111 Cheshire Lane, Suite 500,
Minnetonka, MN 55305

department of Biology, University of Minnesota, Morris, MN 56267

High levels of carbon dioxide are known to be inhibitory to yeast growth, at least at the low temperatures prevailing in the brewing industry, and have also been suggested to favor increased diversion of carbon to glycerol. Since it was not clear whether the inhibitory effects depend on the bulk concentration of CO2 or on its partial pressure, it was not clear whether the same results would be obtained under the higher temperatures employed in fuel alcohol fermentation.

We first determined the conditions prevailing in an industrial com-to — ethanol fermentation plant employing relatively small fermentors, then carried out laboratory fed-batch fermentations with glucose feed with CO2 partial pressures of 0.5, 1.5, 2.5, and 3.5 atm absolute.

Elevated carbon dioxide slowed the fermentation, particularly at the later stages, decreased the maximum number of viable cells obtained and increased cell death rates slightly. High carbon dioxide also decreased overall glycerol production. Low-level aeration also decreased glycerol productivity on a per-cell basis but stimulated cell growth to a compensating extent so that the final level was comparable to the control

Carbon dioxide has both stimulatory and inhibitory effects on the metabolism of living cells. It is known to be required, at relatively low concentrations, by several essential biochemical pathways. For yeast, carbon dioxide concentrations up to 5% in the gas phase have been found to be stimulatory (1,2). Inhibition of various functions begins in at higher concentrations. Aerobic metabolism is significantly inhibited at 0.5 atm CO2 (3), but fermentation per se is not inhibited at 3.5 atm (4) and only begins to be inhibited at 10 atm (5). Anaerobic yeast growth is inhibited

© 1997 American Chemical Society

by lower concentrations, with effects apparent as low as 1.5 atm, depending on the temperature at which the yeast are growing and the strain of yeast (6-9), some selected strains have been propagated at elevated CC>2 concentrations (10). Both rate and extent of growth are affected by inhibitory levels of CO2 under conditions of the brewing industry; the presence of abortive buds and enlargement of the cells suggests interference at specific steps in the cell cycle. Rice et al. present evidence that it is the concentration of dissolved carbon dioxide, and not its partial pressure, which determines the extent of inhibition (6). Thus a given partial pressure of CO2 became less inhibitory as the temperature increased, within the temperature range encountered in brewing. Whether this trend continues into the substantially higher temperature range employed in fuel alcohol production is not known.

The mechanisms involved in CO2 inhibition are unclear, although there are many candidates (11). Carbon dioxide is believed to partition freely into and through biological membranes (12,13), so a purely osmotic mechanism seems unlikely. Yeast cells employ ion-transport mechanisms to maintain their internal pH in spite of the perturbing effect of membrane-permeating weak acids such as CO2, acetic acid, or propionic acid; this is effective in limiting the intracellular pH change to about one unit for a four-unit change in the external pH (14), but carries a cost in energy expenditure. It is not known whether yeast cells have a mechanism for expelling the resulting bicarbonate ion; if they do not, intracellular bicarbonate concentrations could become high enough to inhibit cytoplasmic enzymes (15). Carbon dioxide is similar to other weak acids in inducing potassium uptake by yeast (14). Action at or within the plasma membrane may also account for some or all of the inhibition, similar to mechanisms postulated for ethanol inhibition (16,17).

Oura in 1977 argued that carbon dioxide can also have a substantial influence to increase production of the fermentation coproducts glycerol and succinic acid (18). Glycerol, which is produced to maintain redox balance within the cell, can account for a substantial diversion of carbon away from ethanol production. A major part of the glycerol production occurs to correct a redox imbalance due to production of succinic acid. Decreased carbon dioxide partial pressure and increased available nitrogen were suggested as means of minimizing succinate-associated glycerol production. The available evidence, however, suggests that high carbon dioxide partial pressure decreases rather than increases glycerol production, at least under semi-aerobic conditions in continuous fermentation (19,20).

Although hydrostatic pressures of a few atmospheres have no detectable effects on yeast, hydrostatic pressure increases the saturation concentration of carbon dioxide. Tall fermentors may engender hydrostatic pressures of two atmospheres or more; adding atmospheric pressure and supersaturation may result in local CO2 partial pressures of 3.5 atmospheres near the bottom of a tall industrial fermentor. Indeed, adoption of the use of tall tanks prompted much of the brewing industry’s interest in carbon dioxide effects. We sought to determine whether carbon dioxide effects on fermentation and carbon diversion to glycerol under conditions of the fuel alcohol industry should be a consideration in fermentor design and plant operation.

This investigation had three parts. First, we determined the conditions prevailing during ethanol fermentation in a commercial ethanol plant. Second, we conducted controlled laboratory fermentations at various carbon dioxide pressures. We did not attempt an exact duplication of the industrial process, but rather to simulate certain of the biologically relevant conditions in a more controllable fashion. Thus we used a steady glucose feed to simulate the continuous release of glucose from starch in the industrial fermentation, and used yeast extract to provide the complex nitrogen compounds provided in the industrial proceed by recycled stillage (backset). The remainder of the medium was based on a, well known defined medium to insure nutritional adequacy. Finally, based on what we observed in the laboratory runs, we attempted to decrease carbon dioxide levels and control glycerol production in the industrial process by operation at reduced pressure or with slow air sparge.

Blending conventional DF with esters (usually methyl esters) of vegetable oils is presently the most common form of biodiesel. The most common ratio is 80% conventional diesel fuel and 20% vegetable oil ester (also termed “B20,” indicating the 20% level of biodiesel; see also list of biodiesel demonstration programs in Ref. 6). There have been numerous reports that significant emission reductions are achieved with these blends.

No engine problems were reported in larger-scale tests with, for example, urban bus fleets running on B20. Fuel economy was comparable to DF2, with the consumption of biodiesel blend being only 2-5% higher than that of conventional DF. Another advantage of biodiesel blends is the simplicity of fuel preparation which only requires mixing of the components.

Ester blends have been reported to be stable, for example, a blend of 20% peanut oil with 80% DF did not separate at room temperature over a period of 3 months (126). Stability was also found for 50:50 blends of peanut oil with DF (43).

A few examples from the literature may illustrate the suitability of blends of esters with conventional DF in terms of fuel properties. In transient emission tests on an IDI engine for mining applications (62), the soybean methyl ester used had a CN of 54.7, viscosity 3.05 mm2/s at 40°, and a CP of-2°С. The DF2 used had CN 43.2, viscosity 2.37 mm2/s at 40° and a CP of -21 °С. A 70:30 DF2 : soybean methyl ester blend had CN 49.1, viscosity 2.84 mm2/s at 40°C, and a CP of -17°C. The blend had 4% less power and 4% higher fuel consumption than the DF2, while the neat esters had 9% less power and 13% higher fuel consumption than DF2. Emissions of CO and hydrocarbons as well as other materials were reduced. NOx emissions were not increased here, although higher NOx emissions have been reported for blends (DI engines) (43, 59).

Irregularities compared to other ester blends were observed when using blends of the isopropyl ester of soybean oil with conventional DF (127). Deposits were formed on the injector tips. This was attributed to the isopropyl ester containing 5.2 mole-% monoglyceride which was difficult to separate form the isopropyl ester.

Microemulsification.

The formation of microemulsions (co-solvency) is one of the four potential solutions for solving the problem of vegetable oil viscosity. Microemulsions are defined as transparent, thermodynamically stable colloidal dispersions in which the diameter of the dispersed-phase particles is less than one-fourth the wavelength of visible light. Microemulsion-based fuels are sometimes also termed “hybrid fuels,” although blends of conventional diesel fuel with vegetable oils have also been called hybrid fuels (128). Some of these fuels were tested in engines including the 200 hr EMA test. A microemulsion fuel containing soybean oil, methanol, 2-octanol, and a cetane enhancer was the cheapest vegetable oil-based alternative diesel fuel ever to pass the EMA test.

The components of microemulsions can be conventional DF, vegetable oil, an alcohol, a surfactant, and a cetane improver. Water (from aqueous ethanol) may also be present in order to use lower-proof ethanol (129), thus increasing water tolerance of the microemulsions is important.

Microemulsions are classified as non-ionic or ionic, depending on the surfactant present. Microemulsions containing, for example, a basic nitrogen compound are termed ionic while those consisting, for example, only of a vegetable oil, aqueous ethanol, and another alcohol, such as 1-butanol, are termed non-ionic. Non-ionic microemulsions are often referred to as detergentless microemulsions, indicating the absence of a surfactant.

Viscosity-lowering additives were usually with C,.3alcohols length while longer — chain alcohols and alkylamines served as surfactants. «-Butanol (CN 42) was claimed to be the alcohol most suitable for microemulsions, giving microemulsions more stable and lower in viscosity than those made with methanol or ethanol (130). Microemulsions with hexanol and an ionic surfactant had no major effect on gaseous emissions or efficiency. Emulsions were reported to be suitable as diesel fuels with viscosities close to that of neat DF. No additional engine tests were reported here (130).

Physical property studies of mixtures of TGs with aqueous ethanol and 1-butanol (131) showed that they form detergentless microemulsions. Mixtures of hexadecane, 1-butanol, and 95% ethanol were shown to be detergentless microemulsions. Evidence was presented in that paper that 1-butanol in combination with ethanol associates and interacts with water to form systems exhibiting microemulsion features.

Solubilization and microemulsification studies on TGs, especially triolein, with methanol in the presence of several even-numbered «-alcohols as surfactants showed that 1-octanol produced the microemulsions with the best water tolerance. Among the octanols, 1- and 4-octanol were superior to the 2- and 3- isomers. 1-Butanol and 1- tetradecanol gave microemulsions with the least water tolerance. The formation of molecular dispersions seemed more likely than the formation of nonaqueous microemulsions, but the addition of water produced systems that exhibited microemulsion properties (132). Studies on micellar solubilization of methanol with TGs and 2-octanol as co-surfactant gave the following sequence for water tolerance of three surfactant systems: tetradecyldimethylammonium linoleate > bis(2-ethylhexyl) sodium sulfosuccinate > triethylammonium linoleate. A nonaqueous microemulsion system formed from triolein / oleyl alcohol (9(Z)octadecen-l-ol) / methanol (133).

When studying different unsaturated fatty alcohols, it was reported that the viscosity is nearly independent of the configuration of the double bonds in the tailgroup structure. However, with increasing unsaturation in the tailgroup, viscosity decreased at constant methanol concentration. Generally, adding long-chain fatty alcohols substantially increased methanol solubility in non-aqueous triolein / unsaturated long-chain fatty alcohol / methanol solutions under most conditions. Physical property data were consistent with those for systems exhibiting co-solvent phenomena. However, for solutions with methanol concentration exceeding 0.444 vol frac, the results showed that solubilization of methanol within large aggregates was feasible (134). Mixed amphiphile systems investigating four unsaturated C18 fatty alcohols and five Q — CJ2 alkanols showed that large methanol-in-amphiphile aggregates resembling a microemulsion were feasible under limited conditions (135). These binary systems strongly affect miscibility between methanol and TG. Critical micelle concentration (CMC) studies showed that degree of unsaturation and double bond configuration significantly affected aggregation when using six unsaturated C18 fatty alcohols as amphiphiles (136). These compounds form large and polydisperse aggregates in methanol. The effect of solubilized soybean oil was studied. Viscosity results were consistent with those for microemulsions. Presumably soybean oil is solubilized by incorporation into large soybean oil-in-fatty alcohol aggregates in methanol solvent, resembling a nonaqueous detergentless microemulsion.

Microemulsions containing conventional diesel fuel. Fuel formulations containing conventional DF in emulsion with soybean oil have been subjected to engine testing. In an emulsion with ethanol (737), such a fuel burned faster with higher levels of premixed burning due to longer ignition delays and lower levels of diffusion flame burning than DF, resulting in higher brake thermal efficiencies, cylinder pressures, and rates of pressure rise. NOx and CO emissions increased with these fuels, while smoke and unbumed hydrocarbons decreased. A microemulsion consisting of 50 vol-% DF, 25 vol-% degummed, alkali-refined soybean oil, 5 vol-% 95% aqueous ethanol and 20 vol-% 1-butanol was studied by the 200 hr EMA (Engine Manufacturers Association) test (138). The engine running on this fuel completed the EMA test without difficulty. The microemulsion fuel caused less engine wear than conventional DF but produced greater amounts of carbon and lacquer on the injector tips, intake valves and tops of the cylinder liners besides the observation that engine performance degraded 5% at the end of the test. Another report on blends of alcohols with vegetable oils and conventional DF (the 40:40:20 and 30:40:30 DF/ degummed, dewaxed soybean oil / ethanol blends used in this study were not fully miscible and no surfactant system was used) confirmed that the performance of such fuels was comparable to conventional DF but the tests were too short-term to determine potential problems of carbon buildup, etc. (739).

Microemulsions for blending alcohols with diesel fuel employed unsaturated fatty acids. Saturated fatty acids were unsatisfactory because crystalline phases separated upon refrigeration (129). Addition of V^V-dimethylamino ethanol (DMAE) gave microemulsions with satisfactory viscosity. Two fuels were tested: (1) 66.7% DF2, 16.7% 95% ethanol, 12.5% soybean acids, and 4.1% DMAE (ionic); (2) 66.7% DF2, 11.1% 95% EtOH, and 22.2% 1 — butanol (non-ionic). Both hybrid fuels gave acceptable performance, for example improved brake thermal efficiency and lower exhaust temperatures. Smoke and CO levels were reduced but the unbumed hydrocarbons level increased. The detergentless microemulsion was superior to the ionic one in those SAE properties relevant to good engine performance. On the other hand, fundamental studies on properties of microemulsions such as rheology, density, water tolerance, and critical solution temperatures showed that the water tolerance of ionic systems was greater than that of the 1-butanol system (138). The relative viscosities of the detergentless microemulsion varied directly with the volume percent of the dispersed water phase while for the ionic system the relative viscosities varied with increasing volume percent of dispersed water by values greater than those predicted by theory (140).

Variations of the microemulsion technology have been reported in the patent literature not using vegetable oils but conventional DFs and the fatty ingredient being present only as part of a surfactant system in such emulsions. These microemulsions usually consisted of DF, water, an alcohol (or, combining the latter two components, an aqueous solution of an alcohol), and a system of surfactants. Several such microemulsions with a surfactant system comprising DMAE and a long-chain fatty substance (C9-C22) were patented (141). This microemulsion, which contains a fatty compound only in small amounts, showed a high tolerance for water, which enabled hybridizing diesel fuel with relatively high levels of aqueous alcohol and also showed low-temperature stability. Other systems were a cosurfactant combination of methanol and a fatty acid partially neutralized by a nitrogeneous base such as ammonia, ethanolamine, or /so-propanolamine (142) and, in a similar system, the use of ammonium salts of fatty acids as cosurfactants was patented (143).

Microemulsions with vegetable oils and without conventional DF are the most widely studied. A microemulsion comprising a vegetable oil, a lower (Ci-C3) alcohol, water, and a surfactant system consisting of a trialkylamine or the reaction product of a trialkylamine with a long-chain fatty compound was reported (144). Addition of 1- butanol to the surfactant system was optional. In another patent (145), a microemulsion consisted of a vegetable oil, a CrC3 alcohol, water, and 1-butanol as nonionic surfactant. These fuels had acceptable viscosity and compared favorably to DF2 in terms of engine performance. Another fuel composition consisted of a vegetable oil, methanol or ethanol, a straight-chain isomer of octanol, and optionally water (146), which again had properties such as high water tolerance, acceptable viscosity and performance properties comparable to DF2. Another patent (147) reported the formation of microemulsions from vegetable oil (preferably degummed; mainly rapeseed oil), water, and a surfactant such as an alkaline soap or a potassium salt of fatty acids. Another microemulsion composition was fatty esters, aqueous alcohol, and small amount of alkali metal soap with subsequent separation of the aqueous layer from the microemulsion (148).

Engine tests were performed on several microemulsions. A non-ionic microemulsion comprising of alkali-refined, winterized sunflower oil (53.3 vol-%), 95% aqueous ethanol (13.3 vol-%) and 1-butanol (33.4 vol-%) encountered incomplete combustion at low-load engine operation as major problem (149). Lubricating oil dilution was observed, followed by an abnormal increase in viscosity. Heavier carbon residues on the piston lands, in the piston ring grooves and in the intake ports were noted. Furthermore, premature injection-nozzle deterioration (needle sticking) was experienced. The tested microemulsion was not recommended for long-term use in a

DI engine, but further modifications in formulation might produce acceptable microemulsions.

Two other hybrid fuels were tested. One was non-ionic consisting of 53.3 vol-% soybean oil, 13.3 vol-% 95% aqueous ethanol and 33.4 vol-% 1-butanol (150), and the other was ionic composed of 52.3 vol-% soybean oil, 17.4 vol-% 95% aqueous ethanol,

20.5 vol-% 1-butanol, 6.54 vol-% linoleic acid, and 3.27 vol-% triethylamine. Generally, these fuels performed nearly as well as DF2 despite their lower CNs and less energy content, producing nearly as much engine power (non-ionic emulsion). The increased viscosity of the hybrid fuels produced a 16% increase in the mass of each fuel injection at maximum power, but the injections contained 6% less energy than those of DF2. There was a 6% gain in thermal efficiency.

Another paper reports using methyl tert. — butyl ether (MTBE), which is normally used as octane enhancer in gasoline, to homogenize mixtures of soybean or rape oil with ethanol (151). No engine tests were performed.

In two papers (152-153), emulsions of palm oil with diesel fuel and 5-10% water were tested to determine engine performance and wear characteristics on an IDI diesel engine under steady-state conditions and 20 h endurance tests. Engine performance and fuel consumption were comparable to conventional DF. Wear metal debris accumulation in the crankcase oil was lower than with conventional DF.

Xylitol is produced from D-xylose as a metabolic intermediate in many xylose utilizing microorganisms in two ways: D-xylose is directly converted to xylitol by NADPH- dependent aldehyde reductase (EC 1.1.1.21), or D-xylose is first isomerized to D-

This chapter is not subject to U. S. copyright. Published 1997 American Chemical Society

xylulose by D-xylose isomerase (EC 5.3.1.5) and then reduced to xylitol by NADH — dependent xylitol dehydrogenase (EC 1.1.1.9) (Fig. 1) (<8). Many yeasts and mycelial fungi possess the enzyme xylose reductase which catalyzes the reduction of xylose to xylitol as a first step in xylose metabolism (9). Xylitol production is a relatively common feature among xylose-utilizing yeasts (10). In xylose fermenting yeasts, the initial reactions of xylose metabolism are the major limiting steps (77). This results in the accumulation of xylitol in culture medium, the degree varying with the culture conditions and the yeast strain used (72).

Onishi and Suzuki (73) examined 58 yeast strains belonging to the genera Saccharomyces, Debaryomyces, Pichia, Hansenula, Candida, Torulopsis, Kloeckera, Trichosporon, Cryptococcus, Rhodotorula, Monilia and Torula for polyalcohol production from pentose sugars such as D-xylose, L-arabinose and D-ribose. Candida polymorpha dissimilated aerobically these three pentoses and produced xylitol from xylose, L-arabitol from L-arabinose and ribitol from D-ribose at the yield of 30-40% of sugar consumed. Gong et al. (10) screened 20 strains of Candida belonging to 11 different species, 21 strains of Saccharomyces belonging to 8 species and 8 strains of Schizosaccharomyces pombe for their ability to convert xylose to xylitol. Significant quantities of xylitol were produced by all these yeast strains. Barbosa et al. (14) screened 44 yeasts from five genera (Candida, Hansenula, Kluyveromyces, Pichia and Pachysolen) for conversion of xylose to xylitol. All but two of the strains produced some xylitol with varying rates and yields. The best xylitol producers were localized largely in the species C. guilliermondii and C. tropicalis. Seven strains of C. guilliermondii from diverse isolation sources produced xylitol efficiently when grown in a simple medium containing 5.0% xylose within 24 h (75). However, xylitol essentially disappeared from all the cultures within 72 h. Sirisansaneeyakul et al. (16) selected C. mogii ATCC 18364 as an efficient xylitol producer (Yp/s = 0.62 g/g) from 11 strains of D-xylose utilizing yeasts. Debaryomyces hansenii was an efficient xylitol producer exhibiting a xylitol/ethanol ratio above 4 and a carbon conversion of 54% for xylitol (77). C. entomaea and Pichia guilliermondii produced 0.51 and 0.43 g xylitol/g xylose at pH 5.0 and pH 4.0, respectively and 34°C (18). Ambrosiozyma monospora NRRL Y-1484 produced about 22 g xylitol and 18 g ethanol from 100 g xylose per L when grown at 25°C under moderate aeration (79). A strain of C. tropicalis converted xylose to xylitol and did not produce ethanol (20). Significant quantity of xylitol was produced during ethanol fermentation by Pachysolen tannophilus (21t22) and Kluyveromyces cellobiovorus (23). Various thermo-tolerant yeasts have also been evaluated for the bioconversion of xylose into xylitol (24). Xylitol production ranged from 0.83 to 4.69 g from 10 g xylose.

A fungal strain of Petromyces albertensis produced xylitol when grown in a medium containing D-xylose (25). A large amount (36.8 g/L) of xylitol was obtained from a D-xylose (100 g/L) medium containing ammonium acetate and yeast extract at an initial pH of 7.0. The production of xylitol from xylose has been studied with bacteria such as Enterobacter liquefaciens (26, 27), Corynebacterium sp. (28, 29), and Mycobacterium smegmatis (30).

Onishi and Suzuki (37) screened 128 yeast strains for their ability to produce xylitol from glucose. They reported a sequential fermentation process of xylitol production from glucose (glucose^ D-arabitofcOD-xylulose^xylitol) without isolation

and purification of the intermediates, and the yield of xylitol was 11% from glucose. D. hansenii converted glucose to D-arabitol, Acetobacter suboxydans oxidized D-arabitol almost quantitatively to D-xylulose and C. guilliermondii var. soya reduced D-xylulose to xylitol. Table I summarizes production of xylitol from xylose by some Candida species.

Table L Production of xylitol from xylose by some Candida species

Yeast Fermentation Xylose Xylitol Xylitol

Time (g/L) (g/L) Yield

(h) (g/g)

Factors Affecting Xylitol Production

Medium Components. The conversion of xylose to xylitol by C. guilliermondii was affected by the nutrient source (14). Horitsu et al. (36) studied the influence of culture conditions on xylitol formation by C. tropicalis and optimized the volumetric xylitol production rate by the Box-Wilson method. In this respect, initial xylose concentration, yeast extract concentration and kLa were chosen as independent factors in 23-factorial design. Optimal product formation (г хуШ =2.67 g/L/h, C xytitol =110 g/L) was obtained at 172 g/L xylose, 21 g/L yeast extract and a kLa of 451.5 L/h.

Xylose Concentration. Initial xylose concentration is an important factor to obtain high xylitol production. Meyrial et al. (34) reported that an increase in the initial xylose concentration from 10 g/L to 300 g/L led to activation of xylitol production by C. guilliermondii. The xylitol yield increased gradually with substrate, the highest xylitol yield (0.75 g/g xylose) was obtained at a substrate concentration of 300 g/L. However, the growth of the yeast was gradually inhibited by an increase in initial xylose concentration in the medium. Both the yield and specific rate of cells production declined when xylose concentration initially present in the culture increased. Chen and Gong (32) reported a xylitol yield of 84.5% of theoretical and a maximum production rate of0.269 g/g/h from 249 g/L xylose by Candida sp. B-22. C. tropicalis HXP2 (37) and C. boidinii (33) produced the highest amounts of xylitol (144 g/L and 39 g/L, respectively) at respective values of substrate concentration of 200 g/L and 100 g/L. Dahiya (25) reported maximum xylitol production by P. albertensis was 36.8 g/L at the initial xylose concentration of 100 g/L. Xylitol production declined when the initial xylose concentration was increased to 150 g/L. This might be due to an osmotic effect on cells of P. albertensis or to substrate repression of xylose metabolizing enzymes.

When C. mogii was grown under oxygen-limited conditions in synthetic medium containing different concentrations of xylose (5-53 g/L), the xylitol formation rates showed a hyperbolic dependency on the initial substrate concentration (76).

Vandeska et al. (38) reported that an increase in initial xylose concentration induced xylitol production in C. boidinii but simultaneously acted as a growth inhibitory substrate leading to a long fermentation time. To overcome these problems, fed batch cultures were then used in which higher xylitol yields (0.57-0.68 g/g) and production rates (0.32-0.46 g/L/h) were obtained as compared with a batch process (39). A fed batch process with highest initial xylose concentration (100 g/L) and lowest level of aeration in the first phase, resulted in the highest yield of xylitol (75% of theoretical). A potentiometric biosensor for xylose to monitor fermentative conversion of xylose to xylitol was devised (40).

Presence of Other Sugars. Yahashi et al. (41) investigated the effect of glucose feeding on the production of xylitol from xylose by C. tropicalis. In the bench-scale fermenter (3 L scale) experiment, xylitol was produced at up to 104.5 g/L at 32 h cultivation and a yield of 0.82 (g/g xylose consumed) which is 1.3 times higher than that without glucose feeding. Meyrial et al. (34) evaluated the ability of C. guilliermondii to ferment non-xylose sugars such as glucose, mannose, galactose and L-arabinose commonly found in hemicellulose hydrolyzate. The strain did not convert glucose, mannose and galactose into their corresponding polyalcohol but only to ethanol and cell mass. Arabinose was converted to arabitol. Silva et al. (42) studied batch fermentation of xylose for xylitol production in stirred tank bioreactor. The efficiency of substrate conversion to xylitol was 66% in a medium containing xylose but decreased to 45% in a medium containing xylose and glucose. Vandeska et al. (39) investigated xylitol production by C. boidinii in fed batch fermentations with xylose (50, 100 g/L) and a mixture of glucose (25 g/L) and xylose (25 g/L). All fermentations were initially batch processes with high levels of aeration and rapid production of biomass. Faster growth occurred when a mixture of glucose and xylose was used instead of xylose. Glucose was assimilated first and maximal xylitol production was 39-41 g/L, compared with 46.5 and 59.3 g/L with xylose alone.

Nitrogen Sources and Organic Nutrients. Dahiya (25) studied the effect of 8 ammonium salts and 4 organic nitrogen sources on the production of xylitol from xylose by P. albertensis. Ammonium acetate was most effective for xylitol production. Yeast extract was the most suitable organic nutrient for enhancement of xylitol production. Lu et al. (35) investigated the effect of nitrogen sources [asparagine, casein hydrolyzate, glycine, Traders protein, yeast extract, urea, NaN03, NH 4N03, (NH4) 2S04, NH4C1, NH4H2P04 ] on xylitol production from xylose in shake flasks by an efficient xylitol producing yeast, Candida sp. L-102. Different nitrogen sources influenced xylitol production rate, average specific productivity, and xylitol yield. Maximum xylitol production (100 g/L of xylitol from 114 g/L of xylose) was obtained with urea (3 g/L) as the nitrogen source. Silva et al. (43) evaluated the xylose conversion into xylitol by C. guilliermondii in semi-synthetic media supplemented with different nitrogen sources [urea, NK^Cl, (NH4)2SOJ in a ratio C/N equal 25.6. The type of nitrogen source did not influence this bioconversion and the xylitol yield was around 80%. On the other hand,

Barbosa et al. {14) reported that the use of urea led to higher xylitol productivity by C. guilliermonctii than with ammonium sulfate, and supplementation of urea with casamino acids improved performance over urea alone only slightly. Yeast extract improved yields, but only slightly.

Magnesium and Biotin. Mahler and Guebel {44) studied the influence of Mg+2 concentration on growth, ethanol and xylitol production from xylose by Pichia stipitis NRRL Y-7124. Under constant oxygen uptake rate, biomass/xylose and biomass/Mg+2 yields increased with Mg+2 concentration with a maximum value at 4 mm. Ethanol was the main product formed. At low Mg+2 levels (1 mM), 49% of carbon flux to ethanol was redirected to xylitol production, and was correlated with intracellular accumulation ofNADH.

Lee et al. {45) reported that the relative amount of ethanol and xylitol accumulated in xylose fed aerobic batch cultures of P. tannophilus and C. guilliermondii depended on the limitation by biotin. In high biotin containing media (2 pg/L) P. tannophilus favored ethanol production over that of xylitol while C. guilliermondii favored xylitol formation.

Methanol Supplementation. Dahiya (25) reported that addition of 1% methanol to the medium with 100 g/L xylose increased the xylitol production from 36.8 g/L to 39.8 g/L by P. albertensis. No significant difference in fungal biomass and xylulose accumulation was observed and only 0.015% methanol was consumed. This could be due to the oxidation of methanol to yield NADH which would enhance the reduction of xylose and xylulose to xylitol. Vongsuvanlert and Tani {33) reported about 18 and 26% increase in xylitol production from xylose in presence of 1 and 2% methanol, respectively by C. boidinii. This is also the case with the production of sorbitol from glucose and iditol from L-sorbose by C. boidinii {46).

Initial Cell Density. Cao et al. {47) investigated the effect of cell density on the production of xylitol from xylose by Candida sp. B-22. The rate of xylitol production from xylose increased with increasing yeast cell density. At high initial yeast cell concentration of 26 mg/ml, 210 g/L of xylitol was produced from 260 g/L of xylose after 96 h of incubation with a yield of 81% of the theoretical value. Vandeska et al. {38) reported that high initial cell densities improved xylitol yields and specific production rates of xylitol by C. boidinii. The susceptibility of wood hydrolyzate to fermentation by D. hansenii NRRL Y-7426 was strongly dependent on the initial cell concentration {48).

Oxygen Supply. A variety of yeasts such as Candida, Hanensula, Klyveryomyces, and Pichia require oxygen for sugar uptake {49) and availability of oxygen has significant influence on xylose fermentation by these yeasts {10). However, oxygen limitation is the main factor stimulating the formation of xylitol {50). Roseiro et al (77) reported that xylitol production by D. hansenii required semianaerobic conditions. The presence of oxygen enhanced NADH oxidation and a high NAD+/NADH ratio led to xylitol oxidation to xylulose; therefore, less xylitol was accumulated. Thus the yield of xylitol depended strongly on the oxygen transfer rate (57). Horitsu et al. (56) reported that higher level of dissolved oxygen is required only at the earlier phase of cultivation and afterwards it should be decreased to the lower level of respiration by the yeast. Barbosa et al. {14) reported that increasing oxygen limitation led to increased xylitol productivity and decreased ethanol production with C. guilliermondii. Nolleau et al. (77) evaluated the ability of C. guilliermondii and C. parapsilosis to ferment xylose to xylitol under different oxygen transfer rates. In C. guilliermondii, a maximal xylitol yield of 0.66 g/g was obtained when oxygen transfer rate was 2.2 mmol/1* h. Optimal conditions to produce xylitol by C. parapsilosis (0.75 g/g) arose from cultures at pH 4.75 with 0.4 mmols of oxygen/1* h. The oxygen is not only an important factor to optimize the xylitol production but it is also an essential component for xylose assimilation. When aerobic batch cultures of C. guilliermondii and C. parapsilosis provided with xylose, were shifted to anaerobic conditions, the xylose concentration remained at a constant level and all metabolic activities stopped immediately. C. mogii produced xylitol from xylose under aerobic and oxygen-limiting conditions, but not without oxygen (76). Xylose conversion into xylitol by C. guilliermondii FTI20037 was investigated in a stirred tank bioreactor at different stirring rates {42). Maximal xylitol production (22.2 g/L) was obtained at 30°C, with an aeration rate of 0.46 wm using a stirring rate of 300 per min (lqa = 10.6 h’1). An increase of kLa caused an increase in the consumption of xylose in detriment to xylitol formation. Winkelhausen et al. (52) investigated xylitol formation by C. boidinni in oxygen limited chemostat culture. The production of xylitol by the yeast occurred under conditions of an oxygen limitation at specific oxygen uptake rates lower than 0.91 mmol/gh. The efffect of aeration on xylitol production from xylose by some yeasts is summarized in Table П.

In the traditional jargon of strategic planners, organizations and programs are often called upon to do a "situation analysis" (7). For this type of analysis, planners scan the environment outside and within their organization to identify strengths, weaknesses, opportunities, and threats to it. This is the first step in identifying strategies for success. Strategic planning has clearly lost the allure it once had in this country as an approach to properly setting a course for the future of an organization, but even its most ardent critics

© 1997 American Chemical Society

recognize the value of a situation analysis in formalizing the process of setting strategies

(2).

In the case of bioethanol, we look at the technology itself in identifying strategies for establishing this renewable energy industry. This perspective allows us to look at bioethanol as a new technology for the United States and what it will take to make it a success, while avoiding the more parochial concerns of a given organization. This section is a brief scan of some external issues related to the deployment of bioethanol technology in the United States; it is also an overview of the strategies for deployment currently in place at the U. S. Department of Energy (DOE). External factors that affect the deployment of bioethanol technology include (3):

♦ Environmental (ecological) issues

♦ Energy trends and national security

♦ Public opinion

♦ Public policy and legislative trends

♦ The market.

The Environment. Concerns about the health of our planet and the quality of life that we are leaving for future generations have become a major focus of our society during the past few decades, starting with Rachel Carson’s landmark book Silent Spring (4), to the more recent book Our Stolen Future, which Vice President A1 Gore touted as the unexpected sequel to Silent Spring (5). This literature focuses on the ecological and health effects of manmade chemicals, which travel through the air, water, and soil. A major theme of Coburn’s new book is that even minute levels of many synthetic compounds have unpredictable effects on birth defects and basic human development (5). In this context, "natural” fuel products such as bioethanol look increasingly attractive. The more immediate and quantifiable environmental impacts of bioethanol focus on two key issues: global climate change and urban air pollution.

Global climate change is a surprisingly old issue. The principle of greenhouse gas effects was first proposed by the French mathematician, Fourier, early in the last century (6). In 1896, Svante Arrhenius identified the potential global warming effect of carbon dioxide produced from the burning of fossil fuels (7). But it was not until 1957 that the first definitive proof that carbon dioxide was indeed accumulating in the atmosphere was finally established (8). Many countries (including the United States) have made moves to reduce carbon dioxide emissions, but global climate change remains an issue plagued by political and scientific controversy. Global temperature data show trends of both increasing and decreasing temperature from 1880 to the present (9). Models being developed to predict the effects of increased carbon dioxide levels remain difficult to verify (9). Some members of the scientific community argue that, from a fundamental perspective, such models will never be reliable (10). Even more perplexing is how to predict the effects of a global temperature increase. Regional effects from flooding to droughts have been projected, but clearly we have no way of predicting these types of calamities regionally (10-14). And then there is the question of aerosols. Many have argued that aerosols have similar but opposite effects on climate change compared to greenhouse gases. The presence of anthropogenic aerosols may double the amount of sunlight scattered back into space. The uncertainty in our prediction of aerosol effects swamps any estimate of global warming potential associated with carbon dioxide accumulation.

But if the uncertainties that surround global warming are great, the potential risks to society if global warming is real are worse. Ice core data and other sources of paleoclimatic data have shown that our climate can abruptly and dramatically change (75). These changes have been assumed to occur only during ice ages, but there is now a growing body of evidence that dramatic climate shifts have occurred within the past 10,000 years. These changes are not as dramatic as those observed in glacial periods, but they would still be catastrophic if they occurred today (76). Thus, the comforting notion that periods of warm climate are relatively stable may be incorrect.

And so the controversy continues. Revelle and Seuss’s conclusions on the question of global warming in 1957 still remain the best we can say about the risks we face (5): "Human beings are carrying out a large-scale geophysical experiment of a kind that could not have happened in the past nor be produced in the future. Within a few centuries, we are returning to the atmosphere and the oceans the concentrated organic carbon stored in sedimentary rocks over hundreds of millions of years."

Urban air pollution is another growing environmental problem, especially in terms of its inpact on human health. Nowhere else in the United States has this problem been more evident then in Los Angeles, where 40 years of efforts to control smog and reduce health effects still leave this city with the worst air quality in the United States (77). A wide variety of pollutants may have effects on human health. These include carbon monoxide, sulfur dioxide, and heavy metals. Pollutants, such as nitrogen and sulfur oxides, particulates, and ozone are being scrutinized for their role in respiratory disease. Nitrogen dioxide is one of the most widely recognized respiratory irritants, and it often exceeds safety guidelines in urban areas. There is evidence that short-term spikes in nitrogen oxide are associated with increased hospital admissions for respiratory problems and asthma. Long-term exposure is linked to reduced lung function. In addition, nitrogen oxides contribute to ozone formation. Ozone, formed from the reaction of hydrocarbons, nitrogen oxides, and light, is an important pollutant. Individual responses to ozone vary, but people with respiratory disease or who exercise regularly are particularly prone to reduced lung function caused by ozone exposure (J8).

During the past few decades, the U. S. Environmental Protection Agency’s (EPA) regulation of pollutant emissions from stationary and mobile sources has resulted in major reductions in carbon monoxide, hydrocarbons, and sulfur oxides. Reduction of ozone and smog remains elusive (79). This is, in part, because of the complex atmospheric chemistry involved. EPA’s current strategy for ozone focuses heavily on the reduction of nitrogen oxides. As with global climate change, the problems of dealing with urban ozone are fraught with uncertainty. Developing a more detailed picture of ozone chemistry is critical to developing the most effective strategies for ozone reduction, and to determining the role alternative fuels, such as bioethanol can play in these strategies.

The transportation sector has a major effect on environmental quality, and the use of bioethanol as an alternative to petroleum-based fuel is an important strategy in addressing environmental quality issues. Air pollution, global climate change, oil spills, and toxic waste generation are all results of petroleum-based transportation fuels. The transportation sector contributes almost 30% of the carbon dioxide produced in the United States (20). EPA estimates that transportation contributes 67% of carbon monoxide emissions, 41% of the nitrogen oxide emissions, 51% of reactive hydrocarbon emissions, and 23% of particulate matter emissions (27). DOE estimates that bioethanol use could reduce net carbon dioxide emissions from vehicles by 90% when used as 95% blend with gasoline in light-duty vehicles (22). This is due to the consumption of carbon dioxide by crops used as feedstock for the production of fuel ethanol. Net reductions in urban air pollutants also occur when a 95% ethanol fuel is used. Sulfur oxide emissions are 60% to 80% lower. Volatile organic compound emissions are 13% to 15% lower than those of reformulated gasoline. Net changes in carbon monoxide and nitrogen oxides are marginal (22).

Energy Trends and National Security. Concerns about energy security are among the greatest motivations for the DOE’s Bioethanol Program. It is common sense that some day we will want to switch from a depletable resource such as petroleum to renewable and sustainable sources of energy. The tough questions are how and when. Should we pay more for renewable energy? Do we need to switch sooner rather than later? The answers to these questions have a dramatic impact on the near-term future of bioethanol technology and on how to deploy this technology. Projections for the depletion of domestic sources of conventional and unconventional petroleum suggest that we would run out of domestic oil within 70 years (5). The American Petroleum Institute (API) seems to recognize the legitimacy of these estimates (23), and has used the same U. S. Geological Survey estimates of reserves to show that, if we wished to be absurdly optimistic, there is a 5% probability that we will be able to sustain our petroleum production at current levels for the next 93 years! And there are other ways to extend this sense of optimism. Improvements in technology will increase supply from known resources. Ultimately, API relies on the argument that "unconventional" sources of fuel will triple our resource base. These "unconventional" sources, such as oil shale, are not cost competitive today.

Oil imports are on the rise. The Energy Information Administration estimates that, by the year 2010, we will be importing from 52% to 72% of the oil we consume. Even these estimates may be conservative; in 1995, we crossed the 50% threshold for imported oil. Rising imports not only increase our vulnerability to foreign control of energy supplies (23); they introduce a cost to our economy. DOE’s 2010 projections for imports correspond to economic losses of $114 to $140 billion per year (24). This vulnerability is exacerbated by our transportation sector’s reliance on petroleum for 97% of its fuel demand. These trends show that our energy outlook is clearly becoming a matter of national security.

Public Opinion. Public opinion is one of the most vexing aspects of establishing a strong renewable energy policy. In 1995, a poll asking for people’s priorities in government funding of energy research showed an overwhelming preference for renewable energy as the top priority. But this same poll showed a great deal of ambivalence toward renewables when people use their votes or their pocketbooks to support renewable energy (25). A 1990 report summarizing public opinion polls on energy and the environment during the past 20 years shows erratic swings in our attitudes toward energy security (26). In times of crisis, concern over shortages is high, but it falls off dramatically between events like the Persian Gulf War or periods of long lines at the gas pumps. At the same time Farhar’s report shows a reasonably steady (and more consistent) increase in concern over environmental quality.

Legislation and Policy Debates. Three major pieces of legislation affect the deployment of bioethanol technology:

• The Clean Air Act Amendments of 1990 (CAAA-90)

• The Energy Policy Act of 1992 (EPACT)

• The Alternative Motor Fuels Act of 1988 (AMFA)

CAAA-90 has brought about the increased use of ethanol as an oxygenate in reformulated gasoline for regions considered carbon monoxide and ozone nonattainment areas. EPACT places aggressive mandates on the use of alternative fueled vehicles. The law is intended to force a 10% displacement of fuel consumption with alternative domestic sources by the year 2000. This displacement is to reach 30% by 2010. AMFA complements the efforts of EPACT by putting specific mandates on alternative fuel vehicles for federal fleets. In addition to these three federal laws, there are tax incentives in various states and at the federal level for fuels with a renewable alcohol content of at least 10%. Blenders and sellers of renewable alcohols are eligible for income tax credits. All these incentives are scheduled to expire in the year 2000 (27).

These three pieces of federal legislation attempt to translate environmental and energy security issues into a direct cost of doing business. In other words, these legislative actions force the marketplace to recognize the cost of these societal issues while allowing the marketplace flexibility in finding the most cost-effective solutions to these problems. Vice President A1 Gore has argued that the marketplace has long been blind to many types of "external" costs {28). Even more conservative business-oriented pundits such as Peter Drucker, have come to see this view lacking in the marketplace (29).

The continuing debate in policy circles is how the translation of externalities should be done. In 1992, for example, the Clinton Administration pushed for a "Btu tax" or a "carbon tax" that ultimately lost support and was dropped in favor of a motor fuels excise tax increase that actually penalized alternative fuels (27, 30). Are there uniformly acceptable ways to calculate cost benefits for regulations that address hidden societal costs, such as energy security, the environment, health, and safety? Many economists argue that the answer to this question is yes {31). Such debates become even more difficult when surrounded by a high level of uncertainty. In the case of renewable energy, it is much easier to calculate the cost of mandating their use than to estimate their benefits, because there is no consensus about them. How valuable are reductions in greenhouse gases? Thus, policymakers need to temper their approach with a "risk management" analysis that allows for a balance of probabilities against the seriousness of the risk to society for these externalities {32). European political communities seem more successful at building consensus on these types of issues than their counterparts in the United States {33).

One hundred years after Arrhenius first raised concern about global warming from the burning of fossil fuels, the United Nations’ Intergovernmental Panel on Climate Change (IPCC) concluded that "the balance of evidence suggests that there is a discernible human influence on climate." This may be one of the most important milestones in the policy debate on global warming. That report is now being assailed by an industry group in the United States which argues that no such influence has been proven. Despite the ongoing debate in the United States, the Clinton administration has reversed the U. S. position on limits for greenhouse gases. The United States has now agreed to mandatory limits. Representatives of the U. S. energy industry are reportedly upset by this new position. Specific limits have not been set, and details of how the mandates would be implemented remain unclear (34). Nevertheless, the policy discussion on global warming in the United States has taken one step closer to converting global climate change risk from an externality to real market cost.

The Market. The marketplace will, to the degree that government policy forces it to recognize externalities, ultimately decide the fate of bioethanol as a renewable fuel in the United States. One of the biggest factors that affect this decision is the price of oil. At the low end, DOE projects that oil prices will remain at a level of around $14 per barrel through the year 2010. At the high end, these projections suggest that oil prices could reach $28 per barrel by 2010 (35). DOE has consistently lowered the projected cost of petroleum during the past few years. The same perspective can be seen from annual oil price projections done by IE A (36). With each subsequent year, starting in 1981, IE A pushed its projections for oil prices lower and lower. Prices will arguably remain stable or even drop to levels of $10 per barrel. Stable pricing, so the argument goes, is based only on a psychological momentum that accepts current pricing. But a significant change in the oil market, such as the re-entry of Iraq as a supplier, might bring about a shift down in prices (36).

For the foreseeable future, the market for ethanol will be as an octane enhancer and as an oxygenate in reformulated gasoline (37). The former demand is purely market driven; the latter is driven by CAAA-90 discussed earlier. EPACT and AMFA will have little influence on bringing ethanol into the marketplace as a neat fuel, because its value to a refiner as an octane enhancer and as an oxygenate is much greater than as neat fuel (where it must compete with the continuing low cost of gasoline). A linear programming model has been used to estimate ethanol’s value to refiners. Ironically, ethanol has more value as an octane enhancer than it does as an oxygenate (37). At the lower price projections, for oil, ethanol may demand up to $0.70/gallon. At the high end for oil cost projections, ethanol can compete at around $0.90/gallon. This compares favorably with current estimates for ethanol production in the year 2000 from waste cellulosic materials of around &L08/gallon (Glassner, D., personal communication, 1996). So there is clearly a gap to be filled between the price targets for bioethanol and its market value. The price projection for bioethanol is based on conventional financing of a grassroots (new) ethanol facility: Many other factors could come into play to reduce this cost, including the use of an existing facility to reduce capital cost and unique opportunities for financing and subsidies. Finally, as indicated earlier, the value of ethanol could be influenced by policies that bring externalities involving energy security and environmental quality to bear in the marketplace.

DOE’s Strategy for Bioethanol. DOE outlined its strategies for biofuels in 1994 (38). Reductions in budgets since that time have caused some changes in these goals, but fundamentally, for bioethanol, the goals remain the same:

• Deploy commercial ethanol technology that utilizes waste cellulosic materials by the year 2000

• Introduce the first facility that utilizes a dedicated energy crop, switchgrass, as a feedstock for bioethanol by the year 2005

• Introduce hybrid poplar energy crops for use in bioethanol production after the year 2010

Waste feedstock technology has been chosen for the first deployment, because the feedstocks are cheaper, and such a scenario may be able to take advantage of unique environmental concerns to reduce the cost of ethanol and make it competitive in the market place. Though less predictable, the effect of policies on global warming and environmental quality are expected to improve the value of ethanol above that currently projected in the gasoline market. Clearly, we cannot count on these types of policies, so research and deployment efforts will focus on establishing ethanol production at prices that are supported by the fuel market. These feedstocks include:

• Softwood waste and materials collected from forest-thinning projects in the West aimed at reducing forest fires

• Sugarcane waste

• Hardwood sawdust waste

By 2005, switchgrass should be available for use in a facility that has the technology to use this dedicated energy crop, possibly in combination with other low-cost waste feedstocks. Ultimately, DOE plans to utilize short rotation woody crops, such as hybrid poplars, as a long-term, high-volume resource for producing ethanol. The specific technology approaches for meeting our strategic goals are discussed in the subsequent sections.